Evaluating reanalyses performance in the Baltic Sea region by using assimilated radiosonde data

Abstract

ABSTRACTThe comparison between seven reanalyses – ERA‐Interim, National Centres for Environmental Predictions (NCEP)–National Center for Atmospheric Research reanalysis, NCEP Department of Energy reanalysis, NCEP–Climate Forecast System Reanalysis, Modern‐Era Retrospective Analysis for Research and Application (MERRA), MERRA‐2 and JRA‐55 – and assimilated radiosonde measurements was performed for eight radiosonde stations in the Baltic Sea region during 1980–2015 at five levels. In addition to raw radiosonde measurements, data sets corrected for temperature‐dependence (TD) and time‐lag errors in RS80‐A humidity measurements were investigated. While temperature differences between reanalyses and radiosonde stayed below 0.1 °C at the higher levels, differences up to 1.2 and 2.3 °C were revealed for the 2 m level at 00 UTC and 12 UTC, respectively. At the same time, biases in relative humidity increased with height. The new techniques assimilated in reanalyses were found to be in a disagreement and in unison with radiosonde data, depending on the level, time of the day and reanalysis investigated. The RS80‐A TD correction had a significant effect on tropospheric humidity measurements, increasing specific humidity up to 0.3 g kg−1 at the 2 m level. Based on the ranking results, JRA‐55 is the most accurate reanalysis in representing the radiosonde measurements.